[x265] [arm64] port copy_pp
chen
chenm003 at 163.com
Fri Jul 23 00:18:15 UTC 2021
Looks good, thanks.
2021-07-23 05:52:09,"Pop, Sebastian" <spop at amazon.com>
Hi,
the attached patch ports to arm64 the following kernels:
copy_pp[ 4x4] 16.61x 5.83 96.85
copy_pp[ 8x8] 50.88x 7.41 377.22
[i420] copy_pp[ 4x4] 14.85x 6.30 93.47
[i422] copy_pp[ 4x8] 22.97x 8.22 188.87
copy_pp[ 8x4] 34.93x 5.30 185.06
[i420] copy_pp[ 4x2] 8.28x 5.04 41.73
[i422] copy_pp[ 4x4] 15.72x 6.29 98.81
copy_pp[ 4x8] 23.64x 8.14 192.46
[i420] copy_pp[ 2x4] 6.75x 6.33 42.71
copy_pp[16x16] 66.32x 15.32 1015.63
[i420] copy_pp[ 8x8] 49.04x 7.65 375.08
[i422] copy_pp[ 8x16] 50.99x 14.65 746.82
copy_pp[ 16x8] 97.64x 7.68 750.23
[i420] copy_pp[ 8x4] 31.06x 5.67 176.22
[i422] copy_pp[ 8x8] 48.80x 7.64 372.59
copy_pp[ 8x16] 51.36x 14.72 755.96
[i420] copy_pp[ 4x8] 23.39x 8.45 197.71
[i422] copy_pp[ 4x16] 15.13x 14.94 226.00
copy_pp[ 16x4] 64.05x 5.41 346.35
[i420] copy_pp[ 8x2] 18.09x 4.68 84.73
[i422] copy_pp[ 8x4] 32.38x 5.68 183.86
copy_pp[16x12] 100.95x 11.20 1130.36
[i420] copy_pp[ 8x6] 41.51x 6.66 276.47
[i422] copy_pp[ 8x12] 51.20x 11.00 562.99
copy_pp[ 4x16] 15.67x 14.85 232.74
[i420] copy_pp[ 2x8] 9.44x 8.94 84.34
[i422] copy_pp[ 2x16] 7.96x 22.26 177.21
copy_pp[12x16] 47.22x 23.99 1133.07
[i420] copy_pp[ 6x8] 21.64x 12.94 280.07
[i422] copy_pp[ 6x16] 26.33x 24.38 641.85
copy_pp[32x32] 143.86x 38.77 5577.15
[i420] copy_pp[16x16] 100.09x 14.85 1486.50
[i422] copy_pp[16x32] 96.01x 30.22 2901.08
copy_pp[32x16] 114.02x 21.69 2472.80
[i420] copy_pp[ 16x8] 85.78x 7.86 674.12
[i422] copy_pp[16x16] 91.62x 14.84 1359.77
copy_pp[16x32] 73.04x 30.64 2238.22
[i420] copy_pp[ 8x16] 49.44x 14.62 722.68
[i422] copy_pp[ 8x32] 53.63x 30.08 1613.46
copy_pp[ 32x8] 116.42x 11.64 1355.32
[i420] copy_pp[ 16x4] 58.93x 5.84 344.42
[i422] copy_pp[ 16x8] 88.45x 7.89 697.98
copy_pp[32x24] 135.31x 29.74 4024.18
[i420] copy_pp[16x12] 92.30x 11.27 1039.98
[i422] copy_pp[16x24] 90.90x 22.38 2034.18
copy_pp[ 8x32] 48.70x 30.02 1462.12
[i420] copy_pp[ 4x16] 24.31x 14.94 363.09
[i422] copy_pp[ 4x32] 23.94x 30.43 728.37
copy_pp[24x32] 91.74x 46.34 4251.20
[i420] copy_pp[12x16] 47.35x 24.36 1153.65
[i422] copy_pp[12x32] 43.63x 46.28 2019.02
copy_pp[64x64] 176.98x 129.35 22893.08
[i420] copy_pp[32x32] 144.42x 39.10 5646.94
[i422] copy_pp[32x64] 153.73x 74.20 11406.21
copy_pp[64x32] 171.82x 66.15 11365.43
[i420] copy_pp[32x16] 141.05x 21.61 3048.16
[i422] copy_pp[32x32] 150.24x 38.93 5848.20
copy_pp[32x64] 153.59x 73.28 11254.45
[i420] copy_pp[16x32] 92.93x 30.68 2850.89
[i422] copy_pp[16x64] 95.23x 61.54 5860.84
copy_pp[64x16] 173.45x 35.12 6091.41
[i420] copy_pp[ 32x8] 126.73x 11.92 1510.15
[i422] copy_pp[32x16] 129.62x 21.63 2803.73
copy_pp[64x48] 167.00x 97.67 16311.01
[i420] copy_pp[32x24] 132.80x 30.21 4011.50
[i422] copy_pp[32x48] 154.76x 55.84 8642.34
copy_pp[16x64] 90.05x 61.47 5534.98
[i420] copy_pp[ 8x32] 50.51x 30.05 1517.87
[i422] copy_pp[ 8x64] 48.10x 61.61 2963.15
copy_pp[48x64] 174.26x 98.45 17156.06
[i420] copy_pp[24x32] 90.07x 46.54 4192.25
[i422] copy_pp[24x64] 92.13x 93.10 8577.37
Ok to commit?
Thanks,
Sebastian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20210723/eb86dd42/attachment-0001.html>
More information about the x265-devel
mailing list